What an organ-age result can tell you

In a 2025 blood-protein study of 44,498 people, participants with an extremely older brain profile had about three times the Alzheimer’s hazard of the normal-aging reference group. Hazard describes how new cases occur during follow-up. That is a striking association between groups, rather than a diagnosis or a personal probability of disease. [3]

Organ-age research is helping explain why aging-related risks differ across people and tissues. To interpret an individual report, the important next question is: which measurements and model produced this number, and what decision can it support?

Why these models are interesting

Large studies are finding patterns that may help explain differences in risk between people. Some models have already been tested outside their original research population. The next question is how to use that information to improve care. [1] [2] [3] [4]

How blood can carry information about organs

Blood contains proteins involved in many biological processes. Researchers measure large panels of proteins and use reference data about gene activity in tissues to identify proteins especially associated with particular organs. A statistical model then learns patterns related to calendar age. [1] [3] [4]

An “age gap” is the difference between the model's result and what it expects for someone of the same calendar age. The calculation can be adjusted statistically and expressed in years or in standard deviations—units based on how much values vary in the reference group. An unusual result is defined by the model and its comparison group. [1] [3] [4]

Proteins associated with one organ are not necessarily exclusive to it. Measuring them does not inspect every cell, structure and function of a heart or kidney. Describing the test as simply detecting proteins that “leak from aging organs” makes the method sound more direct than it is. [3] [4]

The useful interpretation is: “This pattern resembles one associated with an older reference group.” “This organ is damaged” is a different claim. So is “this is my fastest-aging organ”: a single age gap does not directly measure change over time. That requires repeated observations and a method suited to interpreting change. [3] [5]

The studies measure different things

Several studies use UK Biobank, so the groups used to develop their models are not wholly independent. Wang and colleagues also evaluated their models in 3,977 China Kadoorie Biobank participants and 800 women from the US Nurses' Health Study. These were selected research samples: the Chinese group came from a heart-disease study and the US group from a colorectal-cancer study. [4]

Testing outside the original population is a useful advance. It does not by itself establish performance in routine care. Wang's paper appeared online in November 2025 and in a 2026 journal issue. The earlier Argentieri study adds a different distinction: predicting whole-body age and later disease does not identify which individual organ needs attention. [2] [4]

How to read the “older brain” finding

The approximately threefold Alzheimer's hazard in Oh's 2025 study compared an extremely older brain profile with the normal-aging reference group. The analysis adjusted for age, sex and variants of the APOE gene. It followed differences between groups; it did not assign treatment or establish an individual's future diagnosis. [3]

A hazard ratio compares how events occur over follow-up. Three times the hazard does not mean three out of ten people will develop Alzheimer's, an increase of three percentage points, or that lowering the score will reduce risk by the same amount. Those would require different information.

The clinical question still depends on the person. Memory symptoms call for assessment of those symptoms. An unusual score in someone without symptoms calls first for an explanation of what the method can and cannot establish. The same score does not automatically create the same next step.

The study also linked multiple extremely older organ profiles with higher mortality. Yet it included a useful counterexample: an extremely youthful artery profile was associated with higher mortality too. The label “younger” was not uniformly favorable across the models. This finding does not mean that youthful arteries themselves are harmful. [3]

Why two organ-age reports can disagree

The organ name is only part of the test's identity. The measurement technology, proteins included, algorithm, reference group and definition of the score all matter. Olink and SomaScan use different technologies to measure proteins. Even a shared protein name does not make the measurement or resulting score interchangeable. A model built on one platform needs suitable validation before being used on another. [3] [4]

Imaging is a more fundamental change of input. Ren and colleagues used 1,777 imaging-derived features across seven organ categories in UK Biobank. The training sample was predominantly White and in middle to later adulthood. Accuracy differed across organs, and independent external validation was still needed. Most participants had imaging at only one visit, so their age gaps did not directly measure individual aging rates. [5]

A blood-protein brain estimate and an MRI-based brain estimate can both carry information while describing different aspects of biology. Likewise, a paper studying thousands of proteins cannot automatically validate a smaller commercial panel of routine blood markers. The smaller panel may be useful, but it needs evidence for its own measurements and purpose. The same requirement applies to a clinic's organ-system assessment, a score based on chemical tags on DNA or a branded organ-age package. [3] [4] [5]

Before comparing two numbers, establish whether they estimate the same thing. Averaging them, choosing the younger result or interpreting disagreement as rapid deterioration adds meaning that those numbers may not support.

Can a treatment change organ-age estimates?

Intervention studies have begun. A 2026 analysis of existing trial data included 185 CALERIE participants from a two-year randomized comparison of calorie restriction and unrestricted eating. It reported smaller increases in several calculated organ- or system-age estimates with calorie restriction. The kidney estimate showed no effect. These were biomarker results, not evidence that using an organ-age report to choose treatment prevents disease or prolongs life. [8]

These reported findings concern the calculations used in that study; they do not establish equivalent effects for the protein and imaging models above. A favorable response in one method cannot be transferred to another. Nor does a result from this analysis establish a personal dietary prescription. [8]

Observational associations with activity, smoking or medication history have a different limit: people with these different histories can differ in many other ways. Such findings can suggest research questions but cannot establish which drug or supplement will make a person's organ healthier. Wang and colleagues identified causal interpretation and intervention assessment as questions still to resolve. [4]

Repeatability, prediction and better care

Three questions need separate answers. Can the method reproduce a result? Does it predict something meaningful, such as disease or loss of function? Does using it lead to better care? A study can succeed at one without settling the other two.

A stable score could add little to care. A powerful predictor could capture risk already apparent from established measurements. A treatment could change a marker without improving how someone feels, functions or survives. The FDA's framework calls a measure used in place of a clinical outcome a surrogate endpoint; it needs substantial evidence in the specific setting where it will be used. An organ-age association alone does not meet that standard. [6]

From a research score to a care decision

The papers discussed here evaluated specific research models. They do not by themselves validate AOSM or its use at HLC. Our published methodology describes how measurements, medical history and physician judgment come together in a care plan. That is a different question from the accuracy of one age estimate.

Imagine a report for a 55-year-old that labels the heart as 63 and the kidneys as 48. These are hypothetical numbers. For a reader hoping to prevent organ disease, the useful distinction is between a model's label and a finding that changes a decision. The heart label does not diagnose heart disease, and the kidney label does not rule out kidney disease. [3] [6]

The clinical conversation should therefore begin with the underlying method, symptoms and established risk measurements. Does the age estimate add information that changes what would otherwise be done? If an abnormal conventional measurement already warrants follow-up, a favorable organ age should not obscure it. For example, US screening guidance for adults without known hypertension connects an elevated office blood-pressure reading to confirmation outside the office before treatment. An organ-age result needs its own justified route from result to action. [7]

The evidence would be stronger if a study showed that decisions guided by a particular organ-age test improved disease or functional outcomes beyond care without that test. That would support a useful role for the test in that type of care. Until that link is established, choosing treatment simply for whichever organ has the largest number is not justified by the report alone. [6]

Read the report before planning another test

Ask for the exact measurement method, model name and version, reference population and study that validates that combination. “Based on Stanford research” or “validated in a large biobank” is too broad when the offered test differs from the research method. Evidence belongs to the exact method and intended use. [3] [4] [5]

Next, ask about uncertainty. Was the method studied in people of your age and health background? Does it give an informative interval or expected variation? A decimal point is a display choice, not evidence of equivalent precision.

Finally, define what a repeat could change. What would follow a higher, lower or unchanged result? Would the proposed action already be appropriate from your history and ordinary assessment? If no result would alter care, another age estimate may add little. The goal is an interpretable finding and a useful next step, rather than the youngest number on the report.

Three questions for a consultation

1. Which exact proteins or images and which model produced this organ age, and was that combination validated in people like me?

2. Does this result add anything to my symptoms and established clinical measurements that would change what we do?

3. If the score changes, how will we tell whether my organ's function or future health has improved?

For more context, explore the guide to useful aging tests and the next steps after receiving results.